Paired evaluation comparison
Compare two systems on the same examples with a reproducible bootstrap interval.
Study guide & lesson outline
LEARNING CONTRACTUnderstand → implement → investigate
Goal: explain the mechanism in this lesson, reproduce one example and test a failure case. Reading alone does not mark mastery.
- Understand: state assumptions, shapes/units and a counterexample.
- Implement: reproduce a numerical example and inspect boundaries.
- Investigate: compare one alternative under a matched budget.
01Configure
02Add your input
Choose settings, load an example, or paste your own input.
03Inspect results
Choose settings, load an example, or paste your own input.
The display may be truncated; the exported JSON report contains all computed results.
How to interpret this
CSV columns: id,a,b. Each ID must occur once and both scores must be finite numbers. Rows must represent the same examples in both systems. The tool resamples paired differences 2,000 times with a fixed seed and reports a 95% percentile bootstrap interval for the mean difference B−A. Assumes independent rows; repeated users or clustered tasks require a cluster-aware analysis. This is not a significance certificate or a power analysis. Limit: 5,000 pairs.
Method references
Assumptions, implementation & acceptance
Resample paired independent seed/run means, not correlated rows from one run. Percentile bootstrap is not a causal claim or a p-value; a degenerate fixture can produce a zero-width interval.
Reviewed counting units, statistical assumptions and numerical examples. Provider accounting and actual GPU allocation still require external measurement.
Before running: predict a boundary case
Resample paired independent seed/run means, not correlated rows from one run. Percentile bootstrap is not a causal claim or a p-value; a degenerate fixture can produce a zero-width interval.
Hint: isolate the assumption
Keep the baseline input fixed, change one parameter or boundary, and compare both the numerical result and the claimed units. A passing example verifies this fixture only.
Minimal executable example
Save the example and linked modules in one folder. Use Node.js 22 or newer, then run node eval-compare.mjs. npm install gpt-tokenizer@4.1.0
import {encode as rawEncode} from 'gpt-tokenizer/encoding/cl100k_base';
import {promptDiff,inspectDataset,chunkDocument,compareScores,estimateVRAM,samplingLab,speculativeLab} from './research-core.mjs';
const encode=s=>rawEncode(s,{disallowedSpecial:new Set()});
const result=compareScores('id,a,b\nx,1,2\ny,4,5\nz,0,1','higher',42);
console.log(result.delta,result.ci); // 1 [1,1]; degenerate fixture, not evidence of zero uncertainty in generalTest and transfer
Resample paired independent seed/run means, not correlated rows from one run. Percentile bootstrap is not a causal claim or a p-value; a degenerate fixture can produce a zero-width interval.
Record input, code version, expected result and observed result in the notebook below. Then construct a counterexample that violates the assumption above and explain why the original conclusion no longer follows.
Download complete kit (.tar.gz) · Complete reference kit and regression checksPredict, test and keep evidence
Before revealing an answer, predict one result. After the exercise, record the discrepancy and a transfer test. Saved locally across language editions.
Open notebook / export / clear →The mathematical review scope is recorded per lesson. It does not establish empirical paper reproduction; report suspected errors in your notes. Review scope